Making Software Predictable by Construction

ABB, a manufacturer of power and automation technologies with headquarters in Sweden, needed a software engineering technology that would ensure that its software and system product lines—which include power, robotics, and process automation—could be customized by third parties while not compromising the governmental and industrial quality standards to which they were built.

ABB’s work with the Predictable Assembly from Certifiable Components (PACC) team at the SEI, which conducts research in achieving predictability by construction, will improve ABB’s ability to quantify its confidence in the quality of its products.

“In short, predictability by construction means taking what you know about the software components of a system—its certified, trusted behavior—and from that predicting how that overall system will behave,” explains Kurt Wallnau, PACC team lead. “What makes our work unique is that we provide the means to have measurable trust in all of this—real, quantifiable, measurable trust.”

“In short, predictability by construction means taking what you know about the software components of a system—its certified, trusted behavior—and from that predicting how that overall system will behave,” explains Kurt Wallnau, PACC team lead.

Partnering with the PACC team will allow ABB to put in place a system to ensure that once a product or service is delivered, any user-customized configurations will hold up to quality tests that ABB had already put in place, explains PACC team member Scott Hissam, who worked on the initiative along with Wallnau and the SEI’s Gabriel Moreno and Dan Plakosh.

The collaboration with ABB began as a result of an SEI research and development project on predictable assembly, which was partially funded by ABB.

In 2006 and 2007, ABB and the SEI applied the key ideas developed in the lab to a release of ABB production software. The intent was to demonstrate, in a product setting, the feasibility and effectiveness of the PACC approach. But first, both teams had to develop a mutual understanding of product and software engineering technology while, at the same time, contending with cultural and language differences.

After several trips by ABB and the SEI across the Atlantic, the ABB engineers—Marcin Stelmarczyk, Saman Hadiani, and Isak Savo—began an extended stay at SEI headquarters to initiate prototype development. The PACC team then constructed a minilab on-site in Pittsburgh replicating the ABB system in Sweden.

The 2006-07 work resulted in a foundation for tools and techniques that can be integrated into ABB’s development environment, architecture, and production software. The results will improve the methods used by ABB to quantify the quality and predictability of its software. This assurance will also improve the methods used by ABB to predict whether a given configuration of systems will violate critical performance requirements.

“The frequent visits, the intense period of travel between the groups, established the personal relationships and expectations. It helps when working remotely,” says Magnus Larsson, manager for industrial software systems at ABB in Vasteras, Sweden.

Although both teams continue to work together, the most beneficial outcome to date has been the development of a system that would allow ABB to predict real-time performance of products. “What we needed was the technology for doing predictions in real-time systems so that we could predict performance behavior,” explains Larsson.

find us here

share this page

Share on Facebook  Send to your Twitter page  Save to del.ico.us  Save to LinkedIn  Digg this  Stumble this page.  Add to Technorati favorites  Save this page on your Google Home Page 

Want more information?

Contact us to find out how the SEI can help.

Government Representative

Al Evans
Manager, Government Program Development
703-908-8225

aevans@sei.cmu.edu

Industry/International Representative

Jan Philpot
Manager, Industry Program Development
703-908-8208

philpot@sei.cmu.edu


Help us improve

Visitor feedback helps us continually improve our site.

Please tell us what you
think with this short
(< 5 minute) survey.